This paper describes a new approach for image-based modeling based on thermographic images and applied to efficiency energy studies of building facades. The approach is automatic and allows obtaining thermographic 3D models and orthoimages, which can be used to assess energy efficiency of building facades. The classical photogrammetry approach is reinforced and combined with computer vision algorithms, especially for the processes of image matching and dense surface generation, so that they have a robust performance with thermographic images. Experimental results presented here demonstrate the capability of the proposed approach using the Polytechnic Advanced School of Avila (Spain) as a case of study. (c) 2013 Elsevier B.V. All rights reserved.